Demis Hassabis, the CEO of Google’s DeepMind, said he’s impressed by DeepSeek’s new AI model but thinks the hype around the Chinese tech is exaggerated.
“I mean, just briefly, on DeepSeek, it’s an impressive piece of work, and I think it’s probably the best work I’ve seen come out of China,” Hassabis said on Sunday in Paris, where he’s set to attend the AI Action Summit.
“But it’s important to understand that despite the hype, there’s no actual new scientific advance there. It’s using known techniques,” added Hassabis. “Actually many of the techniques we invented at Google and at DeepMind.”
He cited Alpha Zero, DeepMind’s learning system that taught itself chess, go, and shogi to the point where it could beat world champions.
Still, Hassabis acknowledged that DeepSeek’s new model could make a difference on the “geopolitical scale.”
The Chinese AI product sent shockwaves through the tech industry last month when it showed that it could rival American models like OpenAI’s ChatGPT, threatening to overturn what the markets had believed was a substantial lead held by US firms.
DeepSeek also said that it spent a fraction of what American firms were investing in their AI models, shaking perceptions of how much it truly costs to develop and improve advanced machine learning. The model’s debut eventually sparked a $1 trillion sell-off in the US market.
Meanwhile, Google also said in its earnings on Tuesday that it’s planning to spend $75 billion on capital expenditure this year.
His team and those at AI powerhouses like OpenAI and Microsoft have been focused on a race to achieve artificial general intelligence, or the point when an AI model can reach or surpass human reasoning.
DeepMind, DeepSeek, and Highflyer, the Chinese hedge fund that backs DeepSeek, did not respond to requests for comment sent by Business Insider.
The post Google’s AI lab CEO said DeepSeek is China’s ‘best work’ he’s seen so far but that it showed ‘no actual new scientific advance’ appeared first on Business Insider.
